Ontology-enabled access control and privacy recommendations

Recent trends in ubiquitous computing target to provide user-controlled servers, providing a single point of access for managing different personal data in different Online Social Networks (OSNs), i.e. profile data and resources from various social interaction services (e.g., LinkedIn, Facebook, etc.). Ideally, personal data should remain independent of the environment, e.g., in order to support flexible migration to new landscapes. Such information interoperability can be achieved by ontology-based information representation and management. In this paper we present achievements and experiences of the di.me project, with respect to access control and privacy preservation in such systems. Special focus is put on privacy issues related to linkability and unwanted information disclosure. These issues could arise for instance when collecting and integrating information of different social contacts and their live streams (e.g., activity status, live posts, etc.). Our approach provides privacy recommendations by leveraging (1) the detection of semantic equivalence between contacts as portrayed in online profiles and (2) NLP techniques for analysing shared live streams. The final results after 3 years are presented and the portability to other environments is shortly discussed.